Good-joinability determining system and good-joinability determining method

Abstract

A tool (22) holds a semiconductor chip (5). A vibrator (25) supplies ultrasonic energy to the tool (22). A sensor (30) acquires a signal waveform of vibration generated between the tool (22) and the semiconductor chip (5) during ultrasonic bonding. An arithmetic portion (15) determines whether or not a bonding state of the semiconductor chip (5) to a substrate (8) is good or not based on the signal waveform.

Description

Technical Field

[0001] This invention relates to a system and method for determining whether an engagement is good or not. Background Technology

[0002] Patent Document 1 discloses an ultrasonic bonding control device, which includes a first vibration sensor that abuts against the component to be bonded, a second vibration sensor disposed on the contact side, and an anomaly detection unit that monitors the output signal of the first vibration sensor or the second vibration sensor.

[0003] In the invention of Patent Document 1, a defective joint, cracking, or abnormal joint peeling is determined by comparing the signal amplitude of the output signal of the first vibration sensor or the second vibration sensor with a first determination threshold or a second determination threshold.

[0004] Existing technical documents Patent documents Patent Document 1: Japanese Patent No. 514873 Summary of the Invention

[0005] However, for example, when multiple bumps provided on a semiconductor chip, which is a first component, are ultrasonically bonded to electrode pads on a substrate, which is a second component, poor bump bonding may occur at a predetermined location within the surface of the semiconductor chip.

[0006] Here, possible causes of poor bump bonding include particles at the bonding interface, deviations in bump height, the stage on which the component being bonded is mounted, and tilting of the semiconductor chip.

[0007] However, in the invention of Patent Document 1, it is only possible to determine whether there is an abnormality in the bonding peeling, cracks, or other defects, and whether the bonding is good or bad. It is not possible to determine whether the bonding state in the plane of the semiconductor chip is good or bad.

[0008] The present invention was made in view of the above-mentioned problems, and its object is to be able to determine whether the engagement state of the first component and the second component is good or not.

[0009] The first invention is a system for determining whether a connection is good or bad. It performs ultrasonic bonding on a first component and a second component and determines whether the bonding state is good or bad. The system comprises: a tool for holding the first component; a vibrator for providing ultrasonic energy to the tool; a sensor for acquiring a signal waveform of vibration generated between the tool and the first component during the ultrasonic bonding process; and a computing unit for determining whether the bonding state between the first component and the second component is good or bad based on the signal waveform.

[0010] In the first invention, during the ultrasonic bonding process, by acquiring the signal waveform of the vibration generated between the tool and the first component, it is possible to determine whether the bonding state between the first component and the second component is good or not.

[0011] Specifically, when the first component and the second component are ultrasonically joined while the first component is held in an inclined position in the tool, the load applied to the joint surface of the first component deviates, which may cause a deviation in the in-plane joint strength of the joint surface of the first component, resulting in poor joint.

[0012] In contrast, in this embodiment, by monitoring and analyzing the signal waveform of the vibration generated between the tool and the first component, specifically the frictional vibration generated by the sliding of the tool and the first component, it is possible to determine whether the engagement state between the first component and the second component is good or not.

[0013] In the second invention, in the system for determining whether the first component and the second component are properly joined, the arithmetic unit determines that the first component and the second component are properly joined when the amplitude of the signal waveform is greater than a predetermined reference value.

[0014] In the second invention, when the amplitude of the signal waveform is greater than a predetermined reference value, it can be estimated that the first component and the second component are normally ultrasonically bonded through metal diffusion, and that more frictional vibration is generated between the tool and the first component. Therefore, it can be determined that the bonding state is good.

[0015] Here, when the first component and the second component are uniformly joined at the joint surface, the reference value can be calculated based on the signal waveform acquired by the sensor.

[0016] In the third invention, within the system for determining whether the connection is good or bad in the second invention, the arithmetic unit acquires the frequency characteristics of the peak and trough components of the signal waveform.

[0017] In the third invention, by analyzing the signal waveform acquired by the sensor into peak and trough components, it is possible to determine whether the engagement state between the first component and the second component is good or not.

[0018] In the fourth invention, in the system for determining whether the connection is good or bad in the third invention, the tool alternately performs a first action of moving towards one end of the vibration direction and a second action of moving towards the other end of the vibration direction during the ultrasonic connection process. The computing unit performs the following processing: based on the change in the current value of the oscillator, it determines whether the tool is in the first action or the second action, synchronizes the first action with one of the peak or trough components of the signal waveform, and synchronizes the second action with the other component.

[0019] In the fourth invention, by synchronizing the first action of the tool with one of the peak or trough components of the signal waveform and synchronizing the second action of the tool with the other component, it is possible to determine whether a poor engagement has occurred on one end or the other end of the vibration direction of the mating surface of the first member.

[0020] The fifth invention is a system for determining whether an engagement is good or bad in any of the first to fourth inventions, wherein a plurality of sensors are provided, and the plurality of sensors are arranged in a position that is linearly symmetrical with respect to a center line extending along the vibration direction of the tool.

[0021] In the fifth invention, by using multiple sensors to monitor and analyze multiple signal waveforms, the accuracy of determining whether the engagement state of the first component and the second component is good or bad can be improved.

[0022] The sixth invention is a method for determining whether a connection is good or bad. The method involves ultrasonically bonding a first component to a second component and determining whether the connection is good or bad. The method includes the following steps: holding the first component by adsorption with a tool, ultrasonically bonding the first component to the second component; acquiring a signal waveform of vibration generated between the tool and the first component during the ultrasonic bonding process; and determining whether the connection between the first component and the second component is good or bad based on the signal waveform.

[0023] In the sixth invention, during the ultrasonic bonding process, by acquiring the signal waveform of the vibration generated between the tool and the first component, it is possible to determine whether the bonding state between the first component and the second component is good or not.

[0024] According to the present invention, it is possible to determine whether the engagement state between the first component and the second component is good or not. Attached Figure Description

[0025] Figure 1 This is a side view showing the outline structure of the system for determining whether the connection is good or bad according to Embodiment 1.

[0026] Figure 2 This is a top view showing the outline structure of the ultrasonic bonding device.

[0027] Figure 3 This is a side view showing the structure of bumps and electrode pads when the semiconductor chip and the substrate are well bonded.

[0028] Figure 4 This is a side view showing a defective bonding condition on the left side of the semiconductor chip's surface.

[0029] Figure 5 This is a side view showing a defective bonding condition on the right side of the semiconductor chip's surface.

[0030] Figure 6 It is a graph showing the signal waveform acquired by the sensor.

[0031] Figure 7 It is a graph showing the signal waveform with peak components.

[0032] Figure 8 It is a graph showing the signal waveform of the trough component.

[0033] Figure 9 This is a flowchart showing the sequence for determining whether an engagement is good or not.

[0034] Figure 10 This is a top view showing the outline structure of the ultrasonic bonding device according to Embodiment 2.

[0035] Figure 11 This is a side view showing a state of poor bonding on the front side of a semiconductor chip within its surface.

[0036] Figure 12 This is a side view showing a state of poor bonding on the rear side of a semiconductor chip within its surface.

[0037] Figure 13 It is a graph showing the signal waveform acquired by the sensor.

[0038] Figure 14 It is a graph showing the signal waveform with peak components.

[0039] Figure 15 It is a graph showing the signal waveform of the trough component.

[0040] Figure 16 This is a flowchart showing the sequence for determining whether an engagement is good or not.

[0041] Figure 17 This is a top view illustrating the configuration of the sensors involved in other embodiments. Detailed Implementation

[0042] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings. However, the present invention is not limited to the following embodiments. Furthermore, appropriate modifications can be made without departing from the scope of the effects achieved by this disclosure. In the figures, arrows indicate directions such as up / down, front / back, left / right. Unless otherwise specified, directions such as up / down will be described according to the directions indicated by these arrows.

[0043] Implementation Method 1 like Figure 1 as well as Figure 2As shown, the ultrasonic bonding device 10 ultrasonically bonds a semiconductor chip 5, which serves as a first component, to a substrate 8, which serves as a second component. At the bonding surface of the semiconductor chip 5 (on... Figure 1 Multiple bumps 6 are provided on the lower surface of the substrate 8. On the bonding surface of the substrate 8 (in...) Figure 1 The upper surface (the middle part) is provided with multiple electrode pads 9.

[0044] The ultrasonic bonding device 10 includes a platform 11, a mounting head drive mechanism 12, a control unit 13, and an ultrasonic head 20.

[0045] A substrate 8 is mounted on platform 11. An ultrasonic head 20 is positioned above platform 11.

[0046] The ultrasonic head 20 is configured to move vertically relative to the stage 11. The ultrasonic head 20 includes a head body 21, a tool 22, and a transducer 25.

[0047] A tool 22 and an oscillator 25 are provided on the head body 21. The tool 22 is located in the head body 21 at a position opposite to the substrate 8 on the stage 11.

[0048] The oscillator 25 delivers ultrasonic energy to the tool 22 via the head body 21. Figure 1 In the example shown, the vibrator 25 is located at the right end of the head body 21. The head body 21 performs ultrasonic vibrations in the left-right direction. The head body 21 functions as an amplitude transformer that amplifies and transmits ultrasonic energy.

[0049] The tool 22, together with the head body 21, performs ultrasonic vibration in the left-right direction. An adsorption hole 23 is formed in the tool 22. The adsorption hole 23 opens at the bottom of the tool 22. The tool 22 adsorbs and holds the semiconductor chip 5 at its lower end by drawing air in through the adsorption hole 23.

[0050] The mounting head drive mechanism 12 moves the ultrasonic head 20 between a mating position and a retracted position. The mating position is where the bumps 6 of the semiconductor chip 5 are pressed against the electrode pads 9 of the substrate 8 for ultrasonic bonding. The retracted position is where the semiconductor chip 5 is removed from the substrate 8.

[0051] The control unit 13 controls the operation of the mounting head drive mechanism 12, the operation of the vibrator 25, and the adsorption operation of the tool 22.

[0052] The ultrasonic bonding device 10 presses the bumps 6 of the semiconductor chip 5 held by the tool 22 toward the electrode pads 9 of the substrate 8 placed on the stage 11 via the mounting head drive mechanism 12, while providing ultrasonic energy to the tool 22 via the vibrator 25.

[0053] Thus, by applying pressure-based loads, heat, and ultrasonic energy between the semiconductor chip 5 and the substrate 8, plastic deformation of the bumps 6 and electrode pads 9 is promoted, resulting in new surface contact between the bumps 6 and electrode pads 9, thereby promoting metal diffusion between the bumps 6 and electrode pads 9. As a result, the bumps 6 of the semiconductor chip 5 are bonded to the electrode pads 9 of the substrate 8.

[0054] <System for Determining Good or Bad Bonding> However, for example, when multiple bumps 6 provided on the semiconductor chip 5 are ultrasonically bonded to the electrode pads 9 of the substrate 8, the bonding strength may be different at a specified position within the surface of the semiconductor chip 5, resulting in poor bump bonding.

[0055] Specifically, during the ultrasonic bonding process, the load applied to the surface of the semiconductor chip 5 may deviate, which may result in different flattening methods of the bumps 6 within the chip surface, leading to differences in the bump bonding strength.

[0056] Here, the causes of poor bump bonding may include, for example, particles attached to the bonding interface between the semiconductor chip 5 and the substrate 8, deviations in the height of the bumps 6 in the semiconductor chip 5, the platform 11 on which the substrate 8 is mounted, and the tilt of the semiconductor chip 5.

[0057] Therefore, in this embodiment, it is possible to determine whether the bonding state between the semiconductor chip 5 and the substrate 8 is good or not.

[0058] Specifically, the bonding success determination system 1 includes an ultrasonic bonding device 10, a sensor 30, and a processing unit 15. The sensor 30 is, for example, an acoustic emission (AE) sensor. The sensor 30 is a semiconductor sensor using a piezoelectric element with a frequency response in the range of several kHz to several hundred MHz. The sensor 30 acquires the signal waveform of the vibration generated between the tool 22 and the semiconductor chip 5 during the ultrasonic bonding process.

[0059] In this embodiment, the ultrasonic energy applied by the transducer 25 is, for example, 60 kHz. However, the applied ultrasonic energy is not limited to this; it can be appropriately set depending on the components being joined.

[0060] Sensor 30 is mounted on ultrasonic head 20. Figure 1 In the example shown, sensor 30 is disposed on the mounting head drive mechanism 12 side of ultrasonic head 20. In this way, by disposing of sensor 30 at a position where the ultrasonic head 20 swings less when ultrasonic energy is applied, monitoring data of the vibration signal waveform can be acquired with high sensitivity.

[0061] Alternatively, sensor 30 can be mounted on stage 11. In this case, the sensor 30 is positioned close to the junction of semiconductor chip 5 and substrate 8, thus reducing noise in the picked-up vibration signal waveform and enabling high-sensitivity acquisition of monitoring data.

[0062] The arithmetic unit 15 determines whether the bonding state between the semiconductor chip 5 and the substrate 8 is good or not based on the signal waveform obtained by the sensor 30.

[0063] like Figure 3 As shown, a good bonding condition is defined as the state in which the bumps 6 in the surface of the semiconductor chip 5 are uniformly bonded to the electrode pads 9 of the substrate 8.

[0064] Here, bump 6 uniformly represents the state in which each bump 6 has a joint strength above the reference under the condition of undergoing a shear test (JIS C 60068-2-21).

[0065] Tool 22 alternately moves towards one end of the vibration direction during the ultrasonic bonding process (in... Figure 1 The first movement (middle is the right side) and the other end (in the direction of vibration) Figure 1 The second movement is the one on the left (the middle position).

[0066] like Figure 4 As shown, during the ultrasonic bonding process, when the semiconductor chip 5 is tilted to the left and downward, the load applied to the bump 6 and the ultrasonic energy become uneven within the chip surface. Therefore, the bonding strength within the surface of the semiconductor chip 5 deviates, resulting in poor bonding. Figure 4 In the example shown, the pressure on the left side of the face of semiconductor chip 5 becomes stronger, while the bonding strength on the right side is insufficient, resulting in poor bonding on the right side of the face of semiconductor chip 5.

[0067] like Figure 5 As shown, during the ultrasonic bonding process, when the semiconductor chip 5 is tilted to the right and downward, the pressure on the right side of the semiconductor chip 5 becomes stronger, while the bonding strength on the left side is insufficient. Therefore, poor bonding occurs on the left side of the semiconductor chip 5.

[0068] <Sensor monitoring of signal waveforms> Figure 6 This is a graph showing the signal waveform acquired by sensor 30. Sensor 30 monitors and acquires signal waveforms representing the frictional vibrations generated between semiconductor chip 5 and tool 22 during ultrasonic bonding.

[0069] Here, frictional vibration refers to the friction generated between the tool 22 holding the semiconductor chip 5 and the semiconductor chip 5 due to sliding after ultrasonic energy is applied to the tool 22 by the oscillator 25, and after the bonding of the bump 6 of the semiconductor chip 5 and the electrode pad 9 of the substrate 8 is completed.

[0070] As signal waveform data acquired by sensor 30, the vibration wave (ultrasonic energy) transmitted from oscillator 25 is observed before ultrasonic bonding.

[0071] As the ultrasonic bonding nears completion, by applying the aforementioned frictional vibration wave, the sinusoidal wave of the transmitted vibration wave changes into a sawtooth wave shape.

[0072] like Figure 7 and Figure 8 As shown, the amplitude of the signal waveform data acquired by sensor 30 is divided into peak and trough components for waveform processing. Waveform processing is then performed to form separate waveforms from the segmented peak and trough components.

[0073] Figure 7 The output waveform is shown after Fourier analysis of the signal waveform with only the peak component. Figure 8 The output waveform is shown after performing Fourier analysis on the signal waveform containing only the trough component.

[0074] Here, the reference value of the output voltage of the sensor 30 obtained from the good bonding that is uniformly bonded in the plane of the semiconductor chip 5 is set as H11.

[0075] The arithmetic unit 15 acquires the frequency characteristics of the peak and trough components of the signal waveform. If the amplitude of the signal waveform is greater than a predetermined reference value, the arithmetic unit 15 determines that the bonding state between the semiconductor chip 5 and the substrate 8 is good.

[0076] Specifically, during the ultrasonic bonding process, if the sensor 30 obtains a signal amplitude H12 that is higher than the reference value H11, it is estimated that the bump 6 and the electrode pad 9 are ultrasonically bonded through metal diffusion, thereby generating more frictional vibration, and it is determined that they are well bonded.

[0077] On the other hand, if the sensor 30 obtains a signal amplitude H13 that is lower than the reference value H11, it is estimated that the bump 6 and the electrode pad 9 failed to achieve ultrasonic bonding through metal diffusion, resulting in less frictional vibration, and is therefore judged as poor bonding.

[0078] Here, tool 22 is oriented towards one end in the direction of vibration (in Figure 1The first movement of the tool 22 (on the right side) is synchronized with either the peak or trough component of the signal waveform. Additionally, the tool 22 is moved towards the other end of the vibration direction (in...). Figure 1 The second movement (left side) is synchronized with the other component of the signal waveform's peak or trough components.

[0079] The arithmetic unit 15 determines whether the tool 22 is in a first operation or a second operation based on the change in the waveform data of the current value of the oscillator 25. In this embodiment, the first operation is synchronized with the peak component of the signal waveform. Furthermore, the second operation is synchronized with the trough component of the signal waveform.

[0080] In this way, by synchronizing the increase and decrease of the current value of the oscillator 25 with the peak and trough components of the sensor 30, the waveform of the peak component is determined as the waveform when the tool 22 is moved to the right by the oscillator 25. Similarly, the waveform of the trough component is determined as the waveform when the tool 22 is moved to the left by the oscillator 25.

[0081] exist Figure 7 and Figure 8 In the example shown, the bonding is successful when tool 22 moves to the right, but fails when tool 22 moves to the left. Therefore, the left side of the face of semiconductor chip 5 is determined to be defective.

[0082] Alternatively, laser Doppler measurement or similar methods can be used to synchronize the left-right swing direction of the actual ultrasonic head 20 with the peak and trough components of the monitoring data from the sensor 30.

[0083] Alternatively, in order to synchronize the left and right swing direction of the ultrasonic head 20 with the peak and trough components of the signal waveform data acquired by the sensor 30, multiple sensors 30 can be configured to predict the poorly joined parts based on the time difference of the acquired waveform data.

[0084] <Regarding the determination of whether the joint is good or not> like Figure 9 As shown, in step S11, the application of ultrasonic waves is started by the vibrator 25, and the process proceeds to step S12.

[0085] In step S12, the change in the current value of the oscillator 25 is obtained, and the signal waveform data of the friction vibration is obtained by the sensor 30, and then the process proceeds to step S13.

[0086] In this embodiment, the time for acquiring the change in the current value of the oscillator 25 and the time for acquiring the signal waveform data by the sensor 30 are set to 200ms after the ultrasonic wave application begins. Furthermore, the acquisition time can be appropriately set based on the chip area, material, number of bumps, bump diameter, bump height, and pin configuration of the joined components.

[0087] In step S13, based on the data representing the change in the current value of the oscillator 25, the left-right oscillation of the tool 22 is synchronized with the peak and trough components of the waveform data from the sensor 30, proceeding to step S14. In this embodiment, the peak component is synchronized with the direction in which the tool 22 moves to the right of the vibration direction. The trough component is synchronized with the direction in which the tool 22 moves to the left of the vibration direction.

[0088] In step S14, waveform processing is performed to segment the signal waveform data into peak and trough components. Furthermore, Fourier analysis is performed on the processed waveform data to identify higher harmonic components compared to the applied ultrasonic frequency. Additionally, based on the properties of sawtooth waves, peaks appear near multiples of the applied frequency; therefore, in this case, peaks are identified around 120kHz and 180kHz.

[0089] If a value higher than the reference value H11 is obtained, the bonding state in the region of that component is determined to be normal. If a value lower than the reference value H11 is obtained, the bonding state in the region of that component is determined to be abnormal.

[0090] If no abnormalities are detected in the waveform processing data (a) of the peak component and the waveform processing data (b) of the trough component, the product is determined to be a good bond in step S15.

[0091] If an anomaly is detected in the waveform processing data (a) of the peak component and no anomaly is detected in the waveform processing data (b) of the trough component, in step S16, it is determined that a poor bonding has occurred in the region on the right side of the inner surface of the chip.

[0092] If no abnormality is detected in the waveform processing data (a) of the peak component and an abnormality is detected in the waveform processing data (b) of the trough component, in step S17, it is determined that a poor bonding has occurred in the region on the left side of the inner surface of the chip.

[0093] If an anomaly is detected in the waveform processing data (a) of the peak component and the waveform processing data (b) of the trough component, in step S18, it is determined that the entire inner surface of the chip is poorly bonded.

[0094] -Effects of Implementation Method 1- As described above, according to the bonding quality determination system 1 of this embodiment, during the ultrasonic bonding process, by monitoring and analyzing the signal waveform of the vibration generated between the tool 22 and the semiconductor chip 5, specifically the frictional vibration generated by the sliding of the tool 22 and the semiconductor chip 5, it is possible to determine whether the bonding state between the semiconductor chip 5 and the substrate 8 is good or not.

[0095] Furthermore, if the amplitude of the signal waveform is greater than a specified reference value, it can be estimated that the semiconductor chip 5 and the substrate 8 are normally ultrasonically bonded through metal diffusion, and more frictional vibrations are generated between the tool 22 and the semiconductor chip 5. Therefore, it can be determined that the bonding state is good.

[0096] In addition, by analyzing the signal waveform acquired by the sensor 30 by dividing it into peak and trough components, it is possible to determine whether the bonding state between the semiconductor chip 5 and the substrate 8 is good or not.

[0097] Furthermore, by synchronizing the first action of the tool 22 with one of the peak or trough components of the signal waveform and synchronizing the second action of the tool 22 with the other component, it is possible to determine whether a poor bonding has occurred on one end or the other end of the vibration direction of the bonding surface of the semiconductor chip 5.

[0098] Implementation Method 2 Hereinafter, the same reference numerals will be used to mark the parts that are the same as in Embodiment 1 above, and only the differences will be explained.

[0099] like Figure 10 As shown, multiple sensors 30 are provided. The sensor 30 includes a first sensor 31 and a second sensor 32.

[0100] The first sensor 31 is disposed on the ultrasonic head 20. When viewed from above, the first sensor 31 is positioned behind the semiconductor chip 5.

[0101] The second sensor 32 is disposed on the ultrasonic head 20. When viewed from above, the second sensor 32 is positioned in front of the semiconductor chip 5. The second sensor 32 is configured to be symmetrical with respect to the center line extending along the vibration direction of the tool 22, i.e., the left-right direction, with respect to the line of the first sensor 31.

[0102] like Figure 11 As shown, during the ultrasonic bonding process, when the semiconductor chip 5 is tilted backward and downward, the load applied to the bump 6 and the ultrasonic energy become uneven within the chip surface. Therefore, the bonding strength within the surface of the semiconductor chip 5 deviates, resulting in poor bonding. Figure 11In the example shown, the pressing force on the rear side of the semiconductor chip 5 in the plane becomes stronger, and the bonding strength on the front side is insufficient, so a poor bonding occurs on the front side of the semiconductor chip 5 in the plane.

[0103] like Figure 12 As shown, during the ultrasonic bonding process, when the semiconductor chip 5 is tilted forward and downward, the pressure on the front side of the semiconductor chip 5 becomes stronger, while the bonding strength on the rear side is insufficient, resulting in poor bonding on the rear side of the semiconductor chip 5.

[0104] <Sensor monitoring of signal waveforms> Figure 13 This is a graph showing the signal waveform acquired by sensor 30. Sensor 30 monitors and acquires signal waveforms representing the frictional vibrations generated between semiconductor chip 5 and tool 22 during ultrasonic bonding.

[0105] As signal waveform data acquired by sensor 30, the vibration wave (ultrasonic energy) transmitted from oscillator 25 is observed before ultrasonic bonding.

[0106] As the ultrasonic bonding nears completion, by applying the aforementioned frictional vibration wave, the sinusoidal wave of the transmitted vibration wave changes into a sawtooth wave shape.

[0107] like Figure 14 and Figure 15 As shown, the amplitude of the signal waveform data acquired by sensor 30 is divided into peak and trough components for waveform processing. Waveform processing is then performed to form separate waveforms from the segmented peak and trough components.

[0108] Figure 14 The output waveform is shown after Fourier analysis of the signal waveform with only the peak component. Figure 15 The output waveform is shown after performing Fourier analysis on the signal waveform containing only the trough component.

[0109] Here, the reference value of the output voltage of the sensor 30 obtained from the good bonding that is uniformly bonded in the plane of the semiconductor chip 5 is set as H21.

[0110] During the ultrasonic bonding process, if the sensor 30 obtains a signal amplitude H22 that is higher than the reference value H21, it is estimated that the bump 6 and the electrode pad 9 are ultrasonically bonded through metal diffusion, thereby generating more frictional vibration, and it is determined that they are well bonded.

[0111] On the other hand, if the sensor 30 obtains a signal amplitude H23 that is lower than the reference value H21, it is estimated that the bump 6 and the electrode pad 9 failed to achieve ultrasonic bonding through metal diffusion, resulting in less frictional vibration, and is therefore judged as poor bonding.

[0112] Here, tool 22 is oriented towards one end in the direction of vibration (in Figure 10 The first movement of the tool 22 (on the right side) is synchronized with either the peak or trough component of the signal waveform. Additionally, the tool 22 is moved towards the other end of the vibration direction (in...). Figure 10 The second movement (left side) is synchronized with the other component of the signal waveform's peak or trough components.

[0113] The arithmetic unit 15 determines whether the tool 22 is in a first operation or a second operation based on the change in the waveform data of the current value of the oscillator 25. In this embodiment, the first operation is synchronized with the peak component of the signal waveform. Furthermore, the second operation is synchronized with the trough component of the signal waveform.

[0114] In this way, by synchronizing the increase and decrease of the current value of the oscillator 25 with the peak and trough components of the sensor 30, the waveform of the peak component is determined as the waveform when the tool 22 is moved to the right by the oscillator 25. Similarly, the waveform of the trough component is determined as the waveform when the tool 22 is moved to the left by the oscillator 25.

[0115] <Regarding the determination of whether the joint is good or not> like Figure 16 As shown, in step S21, the application of ultrasonic waves is started by the vibrator 25, and the process proceeds to step S22.

[0116] In step S22, the change in the current value of the oscillator 25 is obtained, and the signal waveform data accompanying the friction vibration is obtained through the first sensor 31 on the rear side and the second sensor 32 on the front side, and then the process proceeds to step S23.

[0117] In step S23, based on the data representing the change in the current value of the oscillator 25, the left-right oscillation of the tool 22 is synchronized with the peak and trough components of the waveform data from the sensor 30, proceeding to step S24. In this embodiment, the peak component is synchronized with the direction in which the tool 22 moves to the right of the vibration direction, and the trough component is synchronized with the direction in which the tool 22 moves to the left of the vibration direction.

[0118] In step S24, waveform processing is performed to segment the signal waveform data of the first sensor 31 into peak and trough components. Furthermore, Fourier analysis is performed on the processed waveform data to identify higher harmonic components compared to the applied ultrasonic frequency, proceeding to step S25.

[0119] In step S25, waveform processing is performed to segment the signal waveform data from the second sensor 32 into peak and trough components. Furthermore, Fourier analysis is performed on the processed waveform data to identify higher harmonic components compared to the frequency at which the ultrasound is applied.

[0120] If a value higher than the reference value H21 is obtained, the bonding state in the region of that component is determined to be normal. If a value lower than the reference value H21 is obtained, the bonding state in the region of that component is determined to be abnormal.

[0121] If no abnormality is detected in the waveform processing data (a-1) and trough components of the first sensor 31, and the waveform processing data (a-2) and trough components of the second sensor 32, the product is determined to be a good bond in step S26.

[0122] If an anomaly is detected in the waveform processing data (a-1) of the peak component of the first sensor 31, and no anomaly is detected in the waveform processing data (b-1) of the trough component of the first sensor 31, and the waveform processing data (a-2) and trough component of the second sensor 32, respectively, in step S27, it is determined that a poor bonding has occurred in the rear right region within the chip surface.

[0123] If an anomaly is detected in the waveform processing data (b-1) of the trough component of the first sensor 31, and no anomaly is detected in the waveform processing data (a-1) of the peak component of the first sensor 31, and the waveform processing data (a-2) and trough component of the peak component of the second sensor 32, in step S28, it is determined that a poor bonding has occurred in the region on the rear left side of the chip surface.

[0124] If an anomaly is detected in the waveform processing data (a-2) of the peak component of the second sensor 32, and no anomaly is detected in the waveform processing data (a-1) and the waveform processing data (b-1) of the peak component of the first sensor 31, and the waveform processing data (b-2) of the waveform processing data (b-2) of the valley component of the second sensor 32, in step S29, it is determined that a poor bonding has occurred in the front right region within the chip surface.

[0125] If an anomaly is detected in the waveform processing data (b-2) of the trough component of the second sensor 32, and no anomaly is detected in the waveform processing data (a-1) and trough component of the first sensor 31, and the waveform processing data (a-2) of the peak component of the second sensor 32, in step S30, it is determined that a poor bonding has occurred in the region on the left side of the front side inside the chip surface.

[0126] If an anomaly is detected in the waveform processing data (a-1) and trough components of the first sensor 31, and the waveform processing data (a-2) and trough components of the second sensor 32, in step S31, it is determined that the entire inner surface of the chip is poorly bonded.

[0127] Therefore, it is possible to determine the location of a bonding defect occurring within the plane of the semiconductor chip 5. Furthermore, in this embodiment, the data from the first sensor 31 is analyzed first, followed by the data from the second sensor 32; however, the analysis order can be reversed, or the analysis can be performed simultaneously.

[0128] Alternatively, after analyzing the peak components of the first sensor 31 and the second sensor 32, the trough components of the first sensor 31 and the second sensor 32 can be analyzed, or vice versa.

[0129] -Effects of Implementation Method 2- As described above, the bonding condition determination system 1 according to this embodiment can improve the accuracy of determining whether the bonding condition between the semiconductor chip 5 and the substrate 8 is good or bad by using multiple sensors 30 to monitor and analyze multiple signal waveforms.

[0130] Other Implementation Methods The implementation method described above can also be configured as follows.

[0131] In this second embodiment, the structure in which the first sensor 31 and the second sensor 32 are disposed on the ultrasonic head 20 has been described, but the embodiment is not limited to this method. For example, as Figure 17 As shown, the first sensor 31 and the second sensor 32 can also be set on the stage 11.

[0132] In this case, the arrangement is diagonal in both the forward and depth directions relative to the direction of application of the ultrasonic vibration applied from the oscillator. The number of sensors 30 is configured with more than one in each direction. Regarding the configuration position, proximity to the mating semiconductor chip 5 is more preferred.

[0133] Industrial availability As explained above, the present invention is highly practical and useful, as it can determine whether the engagement state between the first component and the second component is good or not. It has high industrial applicability.

[0134] Explanation of reference numerals in the attached figures 1. System for determining whether the connection is good or not 5. Semiconductor chip (first component) 8. Substrate (Second Component) 15. Arithmetic Unit 22 tools 25 oscillators 30 sensors 31 First Sensor 32 Second Sensor

Claims

1. A system for determining whether a joint is good or bad, comprising ultrasonically joining a first component and a second component, and determining whether the joint is good or bad, wherein, The system for determining whether the connection is good or bad has the following features: A tool that holds the first component; An oscillator that provides ultrasonic energy to the tool; A sensor that acquires the signal waveform of the vibration generated between the tool and the first component during ultrasonic bonding; as well as The arithmetic unit determines whether the engagement state between the first component and the second component is good or bad based on the signal waveform.

2. The system for determining whether a connection is good or bad according to claim 1, wherein, When the amplitude of the signal waveform is greater than a predetermined reference value, the arithmetic unit determines that the engagement state between the first component and the second component is good.

3. The system for determining whether an engagement is good or bad according to claim 2, wherein, The arithmetic unit acquires the frequency characteristics of the peak and trough components of the signal waveform.

4. The system for determining whether an engagement is good or bad according to claim 3, wherein, During ultrasonic bonding, the tool alternately performs a first movement toward one end of the vibration direction and a second movement toward the other end of the vibration direction. The arithmetic unit performs the following processing: Based on the change in the current value of the oscillator, it is determined whether the tool is in the first action or the second action. The first action is synchronized with either the peak or trough component of the signal waveform, and the second action is synchronized with the other component.

5. The system for determining whether an engagement is good or bad according to any one of claims 1 to 4, wherein, Multiple sensors are provided. The plurality of sensors are arranged in a position that is linearly symmetrical with respect to a center line extending along the vibration direction of the tool.

6. A method for determining whether a joint is good or bad, comprising ultrasonically bonding a first component to a second component and determining whether the joint is good or bad, wherein, The method for determining whether the joint is good or not includes the following steps: The first component is held in place by adsorption using a tool, and the first component is ultrasonically bonded to the second component. During the ultrasonic bonding process, the signal waveform of the vibration generated between the tool and the first component is acquired; as well as Based on the signal waveform, it is determined whether the engagement state between the first component and the second component is good or not.

Images